LFS-6.6, Stage2, glibc, nscd.c:442

Paul Rogers paulgrogers at fastmail.fm
Wed Jun 2 17:05:36 PDT 2010


> Only one comment: are you *certain* you have removed all related
> things before (or after) restoring a prior-built tarball? Just *one*
> stupid file in the wrong place can bollux the works.

Yes, and yes.  If I had been hand building and retreating by hand, I
agree, I certainly would have been likely to miss something.  That is
why I use package management.  I have used this particular manager
through three previous installations of different LFS versions.  I have
learned how to use it, even upgrade it, and am "certain" I can use it
reliably.  Not to say there are never "user" errors, who can claim that?
I can be certain that the Pass2 gcc was built as properly as can be with
my starting conditions.  It now seems quite clear that the nscd build is
doomed with gcc-3.x, and one must employ a workaround, as distatseful as
that may be to all, myself included.

> > > >  What you are overlooking is that "doing it my way" comes with
> > > >  "when it breaks, I get to keep all the pieces".
> > >
> > > What a curious thing to write in a SUPPORT forum of a LINUX
> > > distribution with the motto, "YOUR DISTRO, YOUR RULES."
> >
> > "Your distro, your rules" - but going along with that, there's the
> > unstated "your rules, your problems". There's nothing wrong with
> > exploring uncharted waters, but if you sail into the middle of "here
> > be dragons", you can't expect to stop and ask for directions...
>
> But one *can* bellow, "Where am I?" And of course, one should then
> expect the obligatory, "Yer in a boat in the midst of the sea, ya
> dang fool!"

But let's not AT ALL slight the purpose of LFS, to teach us how to build
a functional Linux/GNU system.  I'm sure we all agree, it DOES do that.
And if that leads some of us to scoff at the idea of dragons, it has, in
fact, done the job.  I would say that when we instigate workarounds, we
are proving the purpose of LFS.  More power to LFS!

> I don't think you will find any volunteers to perform the intricate
> and time-consuming labor. This is partly because the releases move
> forward fast enough to discourage volunteers.

Perhaps so.  I think it is the responsibility of the development team to
make and preserve HSR reference systems as the base, not the using the
latest prior development version and calling it good.

>   WARNING: some versions of gcc may generate incorrect code with
>   this option.  If problems are observed, a gcc upgrade may be
>   needed.
>
>  and that made me wonder.

Indeed.  As they used a particular source code construct that was
supported by a particular version of gcc, and they did it intentionally,
they might have said, instead of leaving it all mysterious.

> We may never know for sure exactly what caused the original problem
> "undefined reference to __stack_chk_guard" which seems to surface from
> time to time.

Oh, I think linuxfan found the smoking gun somewhere.  It said linux-
2.6.19, gcc-4.2.  OK, maybe not 4.2 if lfs-6.3 works, but at least gcc
on the 4 branch.  The book seems to work for them.

> I haven't followed this thread very closely, so perhaps that explains
> my confusion. What, precisely, is the "problem" with the book? Do you
> want the authors to add a check for the running kernel version? That
> would "fix" the book.

No, the problem with the book is that the Host System Requirements were
incorrect.  Then the error was compounded by (I don't know who) blowing
off these reports of problems building glibc, which seem to have been
indicative, with a "WFM", "unreproducible".  Of course it was
unreproducible if one never had a system with the bare HSR's for
testing.  I dare say if that had been the case, somebody would have
found it reproducible enough.

So, there was never an erratum to the HSR's, until now.

> The purpose of the book I'm sure is multifold, but the primary goals
> are to teach what it takes to build a working minimal Linux system,
> and aid those who, for whatever reasons, find the standard
> distributions not good solutions to their needs, to build a customized
> version of Linux.

And I am living proof that it succeeds in doing that!  Not everyone who
reports a problem here is a fat-fingered newbie, though I once was.  Now
I have hopes, I hope not unfounded, that I can effectively work around
this obstruction by diddling nscd/Makefile to -fno-stack-protector,
compiling glibc, compiling gcc-4.4.3, then rebuilding glibc.  It may not
work with linux-2.6.18.0, but I'll find out before long.

And that's largely if not entirely because I have learned from LFS!

> I don't see that the necessity of using a relatively recent version of
> the kernel in the host environment to be, in spirit, any different
> from requiring a relatively recent version of a C compiler. If all you
> have is a K&R C compiler, then the solution is not to "fix" the book's
> instructions to show how to go about building Linux with such an old
> host. The "fix" is to upgrade the tool set.

Absolutely agree, providing what you tell me is the proper toolset
actually works.  It's not yet clear linux-2.6.18.0 works, although we
are certainly told "Kernel too old" for 2.6.17 with the book's
parameters.

It apparently seems to some that the HSR's are an unimportant
triviality.  Not from the user's point of view.

> My response is a vague sense of being lost in what your goal may be.

I am trying to upgrade from my LFS-6.1 based system, which had been
patched to 2.6.17, then again to .18 to meet or exceed the HSR's, to 6.6
in a consistent upgrade path.  LFS-6.6 is not the sine-qua-non.  That I
can get there from the path I have built going back to LFS-4.1 is
equally important.  The LiveCD approach is not acceptable.  An
intermediate journey from 6.1 through 6.3, barely so, since it is twice
as much work.

> I haven't read where anyone claims that the book is infallible. What
> has been claimed is that if you follow the book, then you'll get help,
> and if you depart then the amount of help you'll get is inversely
> related to the amount of departure.

Cutting and pasting the book text into a script, is not a departure,
IMO.  Surrounding it with additional steps in order to separate the
actual code generation from setup steps, e.g. useradd or the like, or
subsequent steps for adaptation, e.g. editing a fstab, is not a
significant departure.

This is not my first time here.  I know what FBBG means.  Though I
haven't seen the acronym here this time, the attitude still prevails.

> The fundamental problem, as far as I can understand it, is that you
> and the LFS team have different goals.

Doesn't everybody?  Know the story of Procrustes?

> So, the book is "fixed".

There appears to be an appropriate erratum now.  Too bad for all those
who tried to build 6.6 before.  The book should have been checked
against the HSR's it contained BEFORE it was released.

> I'm sure the mail list exists for more than one reason. I always try
> to recall that any help from a list like this is from unpaid
> volunteers. Any help received at all is "pro bono". Complaining that
> free help is not enough sounds uncomfortably like looking a gift horse
> in the mouth.

Up to a point.  If one doesn't want to help people actually SOLVE their
problems, don't put oneself in the line of fire.  They have a problem.
"Well, what we gave you was free!" isn't going to salve their issue.
If one is going to volunteer, go whole hog or not.  Help users with
THEIR problems.  Don't use "volunteer" as an excuse for doing a
half-azzed job.

If you've read "The Cathedral and the Bazaar", I believe one of the
points being made is that FOSS devotees were bringing professional
skills and level of dedication to their products.  That's why FOSS
has succeeded in the face of "daunting competition".  ;-)

> Well, you are the one with the problem building Linux the way you
> want. Others have pointed out a way to do what you want. You don't

Doubling my work by building a 6.3 system first?  Yes.  Is it a wonder
if I might find a workaround like others have found might be appealing?

> Part of what's going on here, I suspect, is that the authors of the
> book foresee lots of work, and not many who need the extra text (like
> perhaps just one).

That's NOT the issue.  As I wrote, the current methodology, avoiding
building from a base system documented in the HSR's, is flawed.  THAT is
the extra work they are avoiding.  Inappropriately, IMO.

> > How far do you suppose we'd have to look to find someone with an old
> > 6.1 system to try the 6.6 book.  Yeah, that's probably too much to
> >   expect, ain't it?
>
> What you seem to be asking is why it's too much to expect that _you_
> figure it out and report back to the list. Only you can answer that
> question.

I thought what I was asking for was recognition that people like me who
have reported this problem repeatedly are in fact the sort of testers
that it was claimed nobody wanted to volunteer for.  True, we are not
interested in trying out extraneous combinations of configurations, but
each of us has had a particular configuration, presumably gcc-3.x in
this case, that has shown a flaw in the book.  Use the resources you are
present with.

> No, it doesn't. There are rules everyone accepts. A poke in the snoot
> is an insult in all cultures. Not all offence has to be taken. Not all
> offences have to be acted upon, however.

And not all criticism should be taken as offensive.  One may not feel it
is justified, but others may.  Did you understand the basis of Pete
Jordan humorous post the other day?

> You've been pointed to a simple way to achieve the goal of building a
> working Linux machine using your hardware as-is, but you won't do it.

Too soon to say.  I won't willingly double my work if there is an
effective workaround.  That's not known yet.  If it doesn't work, then
I'll have to back-up and go another path that may BE more work.  "Unto
the morrow is the evil thereof."

> If your machine is capable of booting off of a CD-ROM, or can boot off
> of a floppy and is capable of reading a CD-ROM, then you can run 6.3
> and do what you want. Does your hardware satisfy

No, I can't.  Look, if having a running system was all that was
important, I could just pop in the LXF DVD I got yesterday and have
KUbuntu-
10.4 running in minutes.  I'm not doing that.  Why?  We have different
goals.  It's important to me to have a consistent path.

> those requirements? I'm sure there are several here who are willing to
> get you going that way. I know I am. I've built a running LFS system
> by doing that.

Without having even looked at the LFS-6.3 book yet, I've learned enough
from LFS installs in the past that I don't anticipate a straightforward
install would be beyond me.  Famous last words, eh? ;-)
-- 
Paul Rogers
paulgrogers at fastmail.fm
http://www.xprt.net/~pgrogers/
Rogers' Second Law: "Everything you do communicates."
(I do not personally endorse any additions after this line. TANSTAAFL :-)

  	

-- 
http://www.fastmail.fm - The professional email service




More information about the lfs-support mailing list